How a Gas Station Owner Cut Stockouts by 40% with AI

When we started working with a fuel delivery business in Taran Taran, Punjab, they had a simple but expensive problem: they kept running out of stock. Not because demand was unpredictable, but because their reordering process was entirely manual. The owner checked tank levels by phone, estimated demand from memory, and placed orders when he felt it was time. Some weeks he ordered too early and tied up capital. Other weeks he ordered too late and lost sales.

The first thing we did was deploy a WhatsApp bot connected to their tank monitoring system. Staff could report dip readings via message, and the AI agent tracked consumption rates automatically. Within a week, we had enough data to start predicting daily demand with reasonable accuracy.

The AI agent now sends a daily WhatsApp message to the owner with a simple summary: current stock levels, projected depletion dates, and a recommended reorder schedule. When stock hits the reorder point, the agent sends an alert with a pre-filled purchase order. The owner approves with a single reply.

After two months, stockouts dropped by 40%. Not because the AI was doing anything magical, but because it removed the human bottleneck from a time-sensitive decision. The owner no longer needed to remember to check stock or calculate reorder quantities. The system did it automatically, every day, without fail.

The financial impact was significant. Each stockout meant lost sales of 15,000 to 30,000 rupees and frustrated regular customers who might take their business elsewhere. Reducing stockouts by 40% translated to roughly 3.5 lakh rupees in recovered monthly revenue.

The lesson here is not about AI sophistication. The forecasting model is relatively simple: exponential smoothing with day-of-week adjustments. The real value is in removing friction from the decision-making process. When the right information reaches the right person at the right time through the right channel (WhatsApp), good decisions happen naturally.